Agriculture Wireless Technology

Document Sample
Agriculture Wireless Technology Powered By Docstoc
					      AGRO-SENSE: PRECISION AGRICULTURE USING SENSOR-BASED
                    WIRELESS MESH NETWORKS
                             Anurag D, Siuli Roy and Somprakash Bandyopadhyay

                                      Indian Institute of Management Calcutta
                                                    Kolkata, India


                       ABSTRACT                                 control of farm equipment is done. The sensing
                                                                technologies allow the identification of pests in the crops,
   Advances in wireless personal area networks have made        drought or increased moisture. These can each have a
the practical deployment of various services possible,          devastating effect on the farm yield. Having such
which until a few years ago was considered extremely            information at a real-time interval, automated actuation
costly or labor intensive. We build such a wireless sensor      devices can be used to control irrigation, fertigation and
network for precision agriculture where real time data of       pest control in order to offset the adverse conditions.
the climatological and other environmental properties are
sensed and relayed to a central repository. The                 The Precision farming system has the following parts:
architecture comprises of three distinct sections – (a) the
sensor-nodes (b) the wireless mesh network and (c) the          a) Sensing agricultural parameters
actuation components. The sensors are selected based on         b) Identification of sensing location and data gathering
the properties suited for the most common crops and we          c) Transferring data from crop field to control station for
identify four such attributes. The sensor network is based         decision making
on the IEEE-802.15.4 standard and we develop a new              d) Actuation and Control decision based on sensed data
static routing algorithm suited for the sensing application.
The algorithm overrides the deficiency of the Hierarchical      Agricultural Sensors, positioning systems for detecting
Routing scheme inherent in the ZigBee specification where       location of sensors, actuators like sprinklers, foggers,
the Cskip addressing algorithm limits the possible depth of     valve-controlled irrigation system, etc. are already available
the network topology due to address wastage. The new            in market. However, very limited work has been done so
algorithm maintains the hierarchical network topology and       far on the technologies to be used to transfer sensor data
thus ensures routing at its optimal best. The algorithms for    wirelessly from crop field to the remote server.
both addressing and routing are provided. The actuation
components are also a part of mesh network and are
activated wirelessly for controlling irrigation and                               2. RELATED WORK
fertigation.
                                                                The concept of precision agriculture has been around for
   Keywords— IEEE 802.15.4, Wireless Mesh Networks,             some time now. Blackmore et al., in 1994 [1] defined it as a
Routing Algorithm, Precision Agriculture                        comprehensive system designed to optimize agricultural
                                                                production by carefully          tailoring soil and crop
                   1. INTRODUCTION                              management to correspond to the unique condition found in
                                                                each field while maintaining environmental quality. The
Precision Farming is the conjunction of a new management        early adopters during that time found precision agriculture
perspective with the new and emerging information and           to be unprofitable and the instances of implementation of
communications technologies leading to higher yields and        precision agriculture were few and far between. Further,
lower costs in the running of large scale commercial            the high initial investment in the form of electronic
agricultural fields. Known also as Site-Specific                equipment for sensing and communication meant that only
Management, precision farming ensures quicker response          large farms could afford it. The technologies proposed at
times to adverse climatic conditions, better quality control    this point comprised of three aspects: (a) Remote Sensing
of the produce and yet a lower labor cost. Precision            (RS), (b) Geosynchronous Positioning System (GPS) and
Farming also makes the use of completely automated              (c) Geographical Information System (GIS). The most
machinery, a possibility. This form of highly automated         important step in precision agriculture is the generation of
agriculture requires intensive sensing of climatic conditions   maps of the soil with its characteristics. These included grid
at the ground level and rapid communication of the raw          soil sampling, yield monitoring and crop scouting. RS
data to a central repository. At the central server, with the   coupled with GPS coordinates produced accurate maps and
availability of computational power, decision making and        models of the agricultural fields. The sampling was
typically through electronic sensors such as soil probes and    •    ZigBee based Wireless Sensor Network with Mesh
remote optical scanners from satellites. The collection of           topology
such data in the form of electronic computer databases gave     •    Very low energy consumption for long battery life
birth to the GIS. Statistical analyses were then conducted
                                                                •    Dynamic, expandable network (up to more than
on the data and the variability of agricultural land with
                                                                     thousands of nodes)
respect to its properties was charted. The technology apart
from being non real-time, involved the use of expensive         •    Able to react on external influences (change
technologies like satellite sensing and was labor intensive          measurement frequency)
where the maps charting the agricultural fields were mostly     •    Different    node   configurations     for    adapted
manually done. Over the last few years, the advancement in           measurement setups
sensing and communication technologies has significantly        •    Measured data stored in a database (easy and variable
brought down the cost of deployment and running of a                 access)
feasible precision agriculture framework. Emerging
wireless technologies with low power needs and low data
                                                                The small catchment area of the Wüstebach (about 26.7 ha)
rate capabilities, which perfectly suites precision
                                                                was proposed to be instrumented with the proposed soil
agriculture, have been developed [2]. The sensing and
                                                                moisture network SoilNet. The sensor network consists of
communication can now be done on a real-time basis
                                                                286 sub nodes and 12 coordinator nodes. The whole
leading to better response times. The wireless sensors are
                                                                network was proposed to be managed by a main server that
cheap enough for wide spread deployment in the form of a
                                                                will also be connected with telecommunication (e.g. DSL)
mesh network and offers robust communication through
                                                                in order to enable online transmission to the workplace.”
redundant propagation paths [3]. Wireless sensor networks
allow faster deployment and installation of various types of
                                                                     3. PROPOSED TEST-BED FRAMEWORK FOR
sensors because many of these networks provide self-
                                                                            PRECISION AGRICULTURE
organizing, self-configuring, self-diagnosing and self-
healing capabilities to the sensor nodes. The applications
using wireless sensor technology for precision agriculture      In this paper, we have designed a Wireless Sensor Network
are briefly explored below.                                     test bed for remote monitoring of agricultural parameters
                                                                and accordingly controlling the irrigation and fertigation
2.1. ‘Smart Fields’ Monitored by Wireless Nano-sensors          leading to Precision Agriculture. In our proposed system,
and the USA’s Plans for a ‘Smart Field System’ [4]              Wireless Sensor Network is realized using IEEE 802.15.4
                                                                based LR-WPAN technology [6]. The architecture can be
“Since many of the conditions that a farmer may want to
                                                                broken into three broad areas:
monitor (e.g., the presence of plant viruses or the level of
soil nutrients) operate at the nano-scale, and because
surfaces can be altered at the nano-scale to bind selectively   1.   Sensing of agricultural parameters
with particular biological proteins, sensors with nano-scale    2.   Addressing of sensors and routing
sensitivity will be particularly important in realizing the     3.   Actuation and control decision based on sensed data
vision of smart fields. The US Department of Agriculture
(USDA) is working to promote and develop a total “Smart         3.1. Description and design of Concept System
Field System” that automatically detects, locates, reports
                                                                The wireless mess network comprises of the sensors placed
and applies water, fertilisers and pesticides - going beyond
                                                                at different locations in a crop field where the intended
sensing to automatic application.”
                                                                characteristics of the soil or atmosphere needs to be
                                                                captured. The sensors’ analogue output is fed into a
2.2. SoilNet - A Zigbee based soil moisture sensor              microprocessor based RF unit like the Chipcon CC2420
network [5]                                                     [7]. The data, now in digital form is packetised and
“Soil moisture plays a key role in partitioning water and       dispatched to the central repository. The path taken by
energy fluxes, in providing moisture to the atmosphere for      packets is determined by a new static routing algorithm
precipitation, and controlling the pattern of groundwater       (explained below). The route based on the location of the
recharge. Large-scale soil moisture variability is driven by    sensor vis-à-vis the central server typically comprises of
space-time precipitation and radiation pattern. At local        multi-hops. Since the sensors with RF modules are static
scales, land cover, soil conditions, and topography act to      and each of them has a unique identity, an apriori
redistribute soil moisture. Despite the importance of soil      association of sensors and its address can be made. The
moisture it is not yet measured in an operational way.”         architecture is pictorially shown in figure 1.

“This project aims to develop a soil moisture sensor            The RF devices can be thought of as active RFID nodes.
network for monitoring soil water content changes at high       Each such node transmits its own identity and its location
spatial and temporal scale. Main features of SoilNet are:       in the form of beacons. Every sensor also has its own
                                                                identity as per its location and this information is apriori
                                                                decided. The RF nodes are required to perform three roles
as provided below. We may thus appropriately design the
capability of the devices.                                            3.3 Addressing of Sensor/Actuator Nodes and Routing
•    Active RFID Gateway                                              The natural choice for the mesh network would be the
•    Active RFID tags with Sensors/Actuators                          ZigBee standard; however, the CSkip algorithm used in
                                                                      ZigBee was found lacking when a sparse, but large network
•    Active RFID Routers
                                                                      is needed. In this section, we first develop the need for tree
                                                                      based network and highlight its properties. We then move
                                                                      on to elicit the deficiencies of the current addressing
                                                                      algorithm in ZigBee. We thus propose a static addressing
                                                                      scheme where the routing is still maintained at its efficient
                                           Remote control Station     best. This algorithm has been tested by way of an
                                          Sensor                      implementation for practical use on TI based boards –
                                          Router                      CC2420 [7].
                                          Gateway
                                                                      3.3.1 Need for a tree based network topology
    Sensors are placed around the zone to be monitored. Routers are   The ZigBee standard based on the IEEE 802.15.4 PHY and
       placed to carry sensor data to central station by multi-hop.
                                                                      MAC Layers provides specifications for two kinds of
                                                                      network topologies- mesh and tree. Mesh networks utilize
          Figure 1. The sensor Network Architecture                   the slightly modified ad-hoc on-demand distance vector
                                                                      (AODV) [8]. However, we are interested and this paper
3.1.1 Active RFID Gateway: Active RFID Gateway is                     delves on the hierarchical or the tree based routing scheme.
essentially the master controller that coordinates the                As said above, the hierarchical scheme is preferred over a
formation of mesh network, collects the tag data and                  mesh networks in the realm of sensing and tracking
transfers it to the host computer. It can also call as                applications. The characteristics of such applications are:
Coordinator. Having a specific role for an RFID
coordinator, where we design and designate a particular               •   A pre-determined area of deployment where the path to
device with the functionalities helps us provide an                       be sensed/tracked is known. Thus the routers (also
abstraction of the inner technology of the mesh network                   known as Full Functional Devices - FFDs) are placed
from the outside world. We may also convey the collected                  along this path.
data over a GSM/CDMA mobile connection and essentially
                                                                      •   The data flow is always between the coordinator and
realize an “Internet of Things” paradigm.
                                                                          an end device (also known as Reduced Functional
                                                                          Device - RFD) and vice-versa. There is never a need
3.1.2 Active RFID tags with Sensor/Actuators: Active                      for two end devices to talk to each other.
RFID tag is connected with sensors can be used as a
sensing device to sense different agro parameters and also
                                                                      The availability of apriori information of the location of the
regulates the actuators as per control decisions.
                                                                      sensors and the coordinator dovetails with the hierarchical
                                                                      tree topology. By design, the location of routers can be
3.1.3 Active RFID routers: Active RFID routers can be                 optimized. This ensures low cost of hardware. The
used for range extension of active RFID devices. Routers              knowledge of the topology also provides the capability of
are fixed at strategic locations within the tracking zones            theoretically determining the fault tolerance capability and
forming a IEEE 802.15.4-compliant wireless mesh network               establishes the bounds on latency. We can also study the
with other routers, gateway and active tags in its vicinity           bounds on network lifetime. Such bounds cannot be
and the location of tags are determined in terms of the               obtained from a mesh topology (since the topology is
locations of fixed routers.                                           random and cannot be predicted). We thus favor the tree
                                                                      topology for planned deployments of sensor networks.
3.2. Sensing of agricultural parameters
Precision Agriculture entails the monitoring of various               3.3.2 Deficiencies of the Cskip Routing Algorithm
parameters which depend on the type of the crop being                 The tree based routing has been shown to be the most
harvested. In our set-up, we have identified four parameters          efficient [9] among the different routing algorithms
which are critical for most types of crops. The parameters            supported by ZigBee. However, this is conditional to the
are:                                                                  successful formation of the network topology in the form of
                                                                      a tree. ZigBee uses the Cskip algorithm for the address
•     Soil pH                                                         allocation and the buildup of the network topology.
•     Soil Moisture                                                   According to the Cskip algorithm, new devices which
                                                                      request association are given a short address. The address
•     Electrical conductivity
                                                                      assigned follows rules set by the network parameters.
•     Soil temperature                                                These parameters are predetermined and are static. The
Parameters are Cm, Rm and Lm; where Cm= Maximum                          deepest router is assigned an address first and the algorithm
Number of children that a full functional device (FFD) can               “works up” the topology. The needs of tracking allow us to
have; Rm = Maximum number or children (out of Cm)                        set the address in advance to preserve the tree structure.
which are FFDs; and Lm = Maximum depth of the network.
The values of these parameters are stored in the NIB                     function main ()
(Network Information Base) in each device [10]. A device                 {
given to a requesting child is generated by the equation:                  assign PAN_Coordinator_address = 0;
                                                                           current_address = En + 1;
         An = A parent + C skip (d ) * Rn + n              (1)             assign_address ( PAN_Coordinator );
                                                                         }

Where An is the address which the new device will take.                  function assign_address ( node )
Aparent is address of the parent of the device that will assign          {
the address. Cskip(d) is determined as follows:                            for ( all children of node from left to right )
                                                                            assign_address ( child_node );

  C skip (d ) =
                  (1 + C   m   − Rm − C m * Rm
                                                 (Cm − d −1)
                                                               )
                                                                   (2)       node_address = current_address;
                                   (1 − Rm )                                 for ( all children of node )
                                                                             {
                                                                               parent_address( child_node ) = current_address;
Cskip(d) determines the block of addresses which the parent                    append into address list, child_node address;
device must skip before assigning the next address. The                      }
algorithm assumes the worst case scenario and makes                          current_address += En + 1;
provision for accommodation for all devices in the pre-                  }
determined network architecture. This assumption of a
worst case scenario severely restricts the network depth.                              Figure 3. The addressing algorithm
For example, values of Cm = 6, Rm = 6 makes Lm=7 for a 16
                                                                         Every Router maintains an address list, which has the
bit short address. The addressing scheme based on Cskip
                                                                         addresses of its child routers. The routing is done based on
develops a tree topology which makes possible the
                                                                         this. The address allocation is pictorially depicted in figure
optimum routing. Routing in such networks is made by
                                                                         5. Assume, a network topology as shown and En = 6. The
comparing the destination address with the Cskip allocation
                                                                         addressing algorithm is akin to the Cskip in the sense of
block. If the destination address is within the Cskip block of
                                                                         addressing in the depth first approach. However, Cskip
any of its children, the packet is forwarded to that child,
                                                                         assumes the worst case and would earmark addresses for
else the packet is forwarded to its parent. It has been shown
                                                                         non-existent nodes. This leads to a huge amount of address
that such tree based routing provides the minimum latency
                                                                         wastage and this precisely is avoided in the static
[9]. The address wastage problem in Cskip is pictorially
                                                                         algorithm. The sensing application has a well defined path
shown below.
                                                                         and this is known in advance. Thus, the need to maintain
                                                                         addresses for future nodes does not arise. Further, the
                       0                                                 address allocation can be suitably tweaked to include an
                                          A sample topology              address provision for future nodes as the need may be. This
          1                          1555 with Rm = Cm = 6. If
                                                                         can be achieved, by including a larger value for En. The
                   Cskip (1)              no more devices join           address of a node denotes the maximum address of all
  2                                       the leg with node “1”,         devices under it. A router assigns address to end devices as
                                          a total of 1551                node_address – n, where 1 <= n <= En. At any given point,
          Cskip (2)            256         (1553 -2) addresses           the property of the node addresses is maintained and thus
                                          are wasted.                    routing is achieved. For example, consider an end device
                                                                         joins the network at node 35. It is assigned an address of
      Figure 2. The address wastage problem in Cskip                     34. The routing decisions would be at each node and
                                                                         checks under which child, can the destination address exist
3.3.2 The Static Addressing Algorithm                                    and forward accordingly. The algorithm is shown in figure
The motivation for a new algorithm arises from the need to               4 and the sample routing in figure 5. The static addressing
prevent the wastage of address space but at the same time                allows for a generic use and supports mobility of the
maintain efficient routing through a tree based structure. A             sensors.
pictorial topology is first prepared with the routers placed
at appropriate distances along the path. The maximum
                                                                         The addressing and routing algorithms have been tested on
number of end devices each router would have to handle is
                                                                         CC2420, TI (Texas Instruments) based board. The network
estimated (En). The address of each router is then
                                                                         layer was written over TIMAC (TI supplied PHY and
determined by the simple algorithm as shown in figure 3.
                                                                         MAC layers).
The algorithm works on the depth first concept where the
address_list = { child router addresses };                               architecture comprises of three distinct components: (a)
if ( dest_address = child_end_device )                                   Intelligent nodes with sensors/actuators (b) The wireless
                                                                         mesh network for communication and the design of a new
  next_hop = dest_address;
                                                                         routing algorithm and (c) The control and actuation. The
else                                                                     sensor design is made based on the parameters to be sensed
{                                                                        for the most common types of crops. We have enlisted four
  next_hop = parent_address;                                             parameters, the soil pH, the Electrical conductivity, the soil
  for all address_list_entries                                           temperature and the soil moisture. The mesh network is
   if ( dest_address of packet <= address_list_entry )                   constructed in the form of a hierarchical tree, but we
                                                                         deviate from the ZigBee standard and build a custom
   {
                                                                         network layer over the PHY and MAC of IEEE 802.15.4.
     next_hop = address_list_entry;                                      The new routing algorithm is based on a static hierarchical
     break;                                                              architecture of the sensors where the need for mobility of
   }                                                                     the sensors is not high. In such networks, we can quickly
  loop                                                                   build a network based on apriori information. In such
}                                                                        scenarios, the address wastage inherent in the Cskip
                                                                         algorithm of ZigBee is avoided. Further, having a
send_packet(next_hop);
                                                                         hierarchical topology ensures routing is maintained at its
                     Figure 4. The Routing Algorithm                     efficient best. The design of the control and actuation is
                                                                         made based on the available hardware in the market. The
3.4. Control and Actuation                                               control decisions work on the sensed parameters through
The actuation is done based on the readings supplied by the              the rules housed in the central repository.
sensors. Upon exceeding a threshold, the system will
generate automated alert messages on the console, upon                                          REFERENCES
which appropriate action can be taken. Automated control
can be incorporated into irrigation and fertigation systems.             [1] Blackmore, S.(1994). Precision Farming: An Introduction.
Currently automated systems are available in drip irrigation                 Outlook on Agriculture 23(4) 4, 275-280.
and foggers. These valve based actuators will be controlled              [2] Ning Wang, Naiqian Zhang, Maohua Wang, “Wireless sensors
through special electrical switches at the controller end.                   in agriculture and food industry—Recent development and
                                                                             future perspective”, published in Computers and Electronics
The actuators (valves) will be operated according to the
                                                                             in Agriculture 50 (2006) 1–14.
control decision. Each actuator will be attached to a RFID               [3] Akyildiz, I.F., Xudong Wang: A Survey on Wireless Mesh
device to identify and control them uniquely.                                Networks, IEEE Communications Magazine (September
                                                                             2005)
                                            address_list =               [4] AZoNanotechnology Article: “Precision Agriculture -
        34 < 56, next_hop = 56         0    {56, 77}                         Nanotech Methods Used, Such as ‘Smart Dust’, Smart Fields’
                                                                             and                                            Nanosensors:”
                                                                             http://www.azonano.com/details.asp.ArticleID=1318
                                                                         [5] “SoilNet - A Zigbee based soil moisture sensor network”
                                 56         77
                                                                             Project Group, Institute Of Chemistry And Dynamics Of The
address_list =
                                                                             Geosphere (ICG), Agrosphere Institute, ICG 4,
{14, 42}
34 > 14,
                                                                             Forschungszentrum       Jülich    GmbH,      52425    Jülich,
                                                       70                    http://www.fz-juelich.de/icg/icg-4/index.php.index=739
34 < 42,                    49
next_hop = 42                                                            [6] IEEE Std. 802.15.4-2003, Wireless Medium Access Control
                                                                             (MAC) and Physical Layer (PHY) Specifications for Low
                                                                    63
                                                                             Rate Wireless Personal Area Networks (2003)
                       14         42                                     [7] CC2420: http://focus.ti.com/docs/prod/folders/print/cc2420.
                                                                             html
                                                  address_ list =        [8] Kwang Koog Lee, Seong Hoon Kim, Yong Soon Choi, Hong
                                                  {34, 28}                   Seong Park: A Mesh Routing Protocol using Cluster Label in
                 7                     35
                                                  next_hop = 34              the ZigBee Network, IEEE International Conference on
                                                                             Mobile Adhoc and Sensor Systems (October 2006)
       An end device joins                                               [9] Ran Peng, Sun Mao-heng, Zou You-min, ZigBee Routing
                                  34                                         Selection Strategy Based on Data Services and Energy-
       the network.                          28               21
                                                                             balanced ZigBee Routing, IEEE Asia-Pacific conference on
 Figure 5. The path of a packet destined to “34” from “0”                    Services Computing (December 2006)
                                                                         [10] Ho-In Jeon, Yeonsoo Kim: Efficient, Real-Time Short
                                                                             Address Allocations for USN Devices Using LAA (Last
                            4. CONCLUSION                                    Address Assigned) Algorithm, 9th International Conference
                                                                             on Advanced Communication Theory (February 2007)
In this paper we have developed architecture for Precision
Agriculture based on wireless sensor networks. The

				
DOCUMENT INFO
Shared By:
Categories:
Stats:
views:91
posted:12/9/2010
language:English
pages:5
Description: Agriculture Wireless Technology document sample